My experience is that the universe designer creates the connection, and the supervisor maintains it. Although, this may be the same person, in some deployments.
I usually give the co0nnection the same name as the universe that uses it.
Hmmm, I can’t answer these. 
I prefer “disconnect after each transaction”. This seems to minimize the chance of running out of connections to the database. Other people may have other thoughts.[/quote]
MichaelWelter 
(BOB member since 2002-08-08)
Q1) In our scenario (various departments, various DBs) we decided to only allow the Gen Sup to create connections. Designers started to create all kind of connections, and these connections have to be maintained on the ZABO server as well (DB client middleware) - therefore centralized creation of universe connections.
Andreas 
(BOB member since 2002-06-20)
In our situation, there are just two of us who are both designers/supervisors/trainers/user support, so our answers are:
Anita Craig 
(BOB member since 2002-06-17)
Changes with every location I go to. 
I saw the following once and found it to be real good…
avaksi (BOB member since 2002-08-22)
A few designers and supervisor who know what they are doing
Not as much as we should! 
I’ve seen one of your BCA error posts and I don’t think this will help you :?
The only thing I have ever used it for is to add an oracle hint. This applies to ALL queries run on the universe using that connection
I’d go with the others on this - disconnect after each transaction. Either Steve or Dave wrote something that said because the time taken to actually logon to databases is now so quick now it makes sense to chose this setting
Nick Daniels 
(BOB member since 2002-08-15)
I don’t know that I’ve done it correctly, but it made sense to me for ease of maintenance to create one connection for each server. So I have one connection each for Production, QA, and Development. That connection has access to all of the tables in each environment. I am the only one who has created connections, although I’ve given all of the IT developers rights to do so.
My naming convention as also done to make it easy. Our department is HEARS. I’ve got HEARS_PROD, HEARS_QA, and HEARS_DEV.
Eileen King (BOB member since 2002-07-10)
Sounds like we aren’t too far off in what we are doing. We’ll test changing the connection properties to Disconnect after each transaction.
Thanks everyone for all of the information!
Sandy Brotje (BOB member since 2002-07-24)
It allows you to set specific options for your database. For example, you can set up a “Star” hint when using Oracle to tell the optimizer you are using a star schema. Oracle will then optimize your queries differently. You can set up initialization options for databases that support them. Basically, anything that doesn’t fit on the first two tabs that is specific to your flavor of database goes here. 
And the connection info really has nothing to do with that. Keep in mind that there are two typical reasons given for why ZABO is slower than full client. First, there are 3 machines (client -> Webi server -> database) instead of two (client -> database). That adds more network traffic, and more conjestion at the Webi server. So that slows things down. Second, ZABO communicates with the Webi server via HTTP, which was not really desiged for this sort of work. It works, but it wasn’t designed for massive amounts of data.
If you have ever downloaded a file via HTTP and downloaded the equivalent file via FTP, you probably noticed that the FTP transfer completed much faster than the HTTP. Why? Because it (the FTP protocol) was designed for that sort of activity.
Dave
Dave Rathbun (BOB member since 2002-06-06)
I agree. I experimented with keeping my connection open longer in an effort to get rid of those updating messages we’re discussing in the other thread but it didn’t help. It may have gotten rid of some connecting but you said you only saw one of those today.
I always found that Zabo was slower that connecting without the web server but not the huge difference you saw. But, it may be that my server was better, stronger, faster, and closer.
My main concern is those updating messages once for each data provider. We get rid of those and we see a faster prompt display. At least that was my experience!
Cindy Clayton (BOB member since 2002-06-11)
excellent topic and i have a question about this too but 1st…
now other questions…
what about synchronous vs async? i understand that async is good if only to allow the user to minimize bo while they are running a query on their desktop BUT as we have a webi deployment, per the 513 readme, you cannot use async connections on webi.
is it important to keep your universe connections in-line with you bomain.key connection?
csievert 
(BOB member since 2002-06-17)
Well, you can use async connections for your WEBI server, but it is not supported (we ran into problems using async connections regarding refreshing LOVs and had to switch to sync connections for WEBi, that was back on 2.5.x I think).
I am curious: Does anyone know the exact details why WEBi on async connections is not supported/not working properly and if this will change in the future?
Andreas (BOB member since 2002-06-20)
My explanation - and it is conjecture (fancy word for “guess”) - is that it is because HTTP is essentially an asynchronous connections. Think about it… when you connect to a web site, you can interrupt it, stop it, do whatever you want. So to keep the database connection in sync with your webi session, it requires that the back end, at least, be synchronous.
Also, an async connection can be interrupted at any time… and it’s difficult to know whether the connection was interrupted or wheter it was actually finished. By requiring a synchronous connection, when the data stops coming, we assume that it’s all there. Again, this is all just a guess.
Even with synchronous connections, you should be able to minimize your webi session and go on with other tasks.
And to answer your last question… unless they change the protocol used to transfer information, I don’t expect this to change.
Dave
Dave Rathbun (BOB member since 2002-06-06)
just to be clear (in case this was in reference to my user issue about minimizing) my users wants to be minimizing full client, not webi. this is simple on win2k or nt because of the ‘show desktop’ but not 98.
csievert (BOB member since 2002-06-17)
I think the most successful connections I’ve made are the ones I made at the BO User Conference.
MichaelWelter (BOB member since 2002-08-08)
preach it! 
csievert (BOB member since 2002-06-17)
I have to agree…and I learned more about some interesting things you can do with your connections in Business Objects too!
Eileen King (BOB member since 2002-07-10)
Dave,
I have made a hint in a connection pointing to oracle 8.1.7 with /*+ STAR_TRANSFORMATION */. Unfortunately the connection did not work anymore and in the SQL I could not see the hint. Do you have a clue, what I have done wrong?
To work around, I have created in the universe an additional object with the value /*+ STAR_TRANSFORMATION */ user. If I add this object in every report the hint is in the sql, but I could not yet test whether the access path is better or not.
Cheers
Urs
Urs Aeschbacher (BOB member since 2002-09-02)
What is the proper syntax for including a database hint for Oracle? Should the hint appear in the report SQL or only in the SQL that runs in the database?
Thanks,
Kim
kim snead (BOB member since 2002-08-21)
Example for universe object using an Oracle hint (in the SELECT box):
/*+RULE*/ ' 'This universe hint object must be the first object in the result set when building a data provider, and it should show up in the Business Objects generated SQL code.
Andreas (BOB member since 2002-06-20)
Is this the same syntax that you would include when adding the hint to the universe connection using the “Custom” tab from the “edit” connection option?
kim snead (BOB member since 2002-08-21)